Persian MusicGen model fine-tuned with large dataset for cultural music generation

Researchers have developed a culturally-aware generative music model for Persian music, named Persian MusicGen. This model was fine-tuned using a newly curated dataset of over 900 hours of Persian audio samples, covering various genres. The fine-tuned model demonstrates improved adherence to Persian stylistic conventions compared to models trained on Western music. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Enables more culturally relevant AI music generation, potentially opening doors for other niche musical styles.

RANK_REASON Publication of an academic paper introducing a new dataset and a fine-tuned generative model. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• Persian MusicGen
• MusicGen
• Mohammad Hossein Sameti

• paper
• model release